Salut,
On 10-04-26 9:02 AM, Oleksiy Stashok wrote:
>> OK cool, so can you trace the code to see what going on?
>>
> Yes, I've even described what I see :)) There is no change in code path,
> except one, which is called just once.
> It happens *before* login, when browser's script initializes async
> atmosphere connection. Atmosphere doesn't set Content-Type header for
> the HTTP response, see AtmosphereResourceImpl.write(), so the
> Content-Type header is null, and in this case Grizzly sets the default one.
>
> Otherwise there is no login/post or any other issues with the
> applications. In the wireshark I see those messages are sending/coming
> absolutely the same way as before.
>
> IMO, either browser or specific java script doesn't accept content-type
> header for the Atmosphere async. HTTP connection.
>
> The simple fix for Atmosphere (in order to make it work on Grizzly) is
> to add following line into AtmosphereResourceImpl.write():
>
> **********
> void write() {
> // Set to expire far in the past.
> res.setHeader("Expires", "-1");
> // Set standard HTTP/1.1 no-cache headers.
> res.setHeader("Cache-Control", "no-store, no-cache, must-revalidate");
> // Set standard HTTP/1.0 no-cache header.
> res.setHeader("Pragma", "no-cache");
> ++++ res.setHeader("Content-Type", ""); <------------- This one
We can't set that value as it will breaks the Jersey module and the cpr
as well. Other WebServer aren't doing that (setting the default when
unspecified)...I think it should be configurable in Grizzly, e.g allow
framework to disable it and allow response without content-type.
WDYT?
A+
-- jeanfrancois
> ***********
>
> May be it's possible to fix this directly on the client side, don't know.
>
> WBR,
> Alexey.
>
>>
>>> On Apr 26, 2010 7:06 AM, "Oleksiy Stashok" <Oleksiy.Stashok_at_sun.com
>>> <mailto:Oleksiy.Stashok_at_sun.com>> wrote:
>>>
>>> Sebastien,
>>>
>>> you didn't understand me (may be I wasn't clear), I did run the test
>>> and did reproduce the issue! And agree that it's not working now and
>>> that for rev. < 4420 it worked.
>>>
>>> In my comments I wrote about observations, what the rev. 4420 changed
>>> for the test. The only change I see is the content-type header I
>>> wrote about.
>>>
>>> WBR,
>>> Alexey.
>>>
>>>
>>>
>>>
>>> On Apr 26, 2010, at 12:28 , Sebastien Dionne wrote:
>>>
>>> > please. do the test after the login... the...
>>>
>>
>